Three Evaluators. Two Metadata Pipelines. 3.5 Hours of Clinical Contact. One Verdict.

The Rule 20.01 competency machinery was ordered four days after the arrest. Twenty days before Guertin’s patent was granted. At that point the case docket contained 8 filed documents totaling 16 pages. The 597 DKIM-authenticated LinkedIn searches had not yet been catalogued. The 2.9 million-row forensic database did not exist. The framework that would classify the claims as symptoms was already in motion before the evidence had finished accumulating.

Three evaluators would examine him over the next twenty-two months. Rogstad (March 2023). Milz (January 2024). Cranbrook (December 2024). All three arrived at the same diagnosis. All three worked for the Fourth Judicial District Psychological Services — the same organizational unit, reporting to the same Chief. None verified a single factual claim the defendant presented. None entered a patent number into a free federal database. None acknowledged, in their reports or opinions, that the Minnetonka Police Department incident report Guertin filed nine days before the arrest — case number MP23000151, January 12, 2023 — existed as an independent documentary record of his patent-theft concerns. Rogstad’s own report lists that police report under “records reviewed.” One conducted zero hours of clinical contact.

What follows is what each report actually says. Every quote is verbatim. Every source is linked.

Rule 20 as Foregone Conclusion

January 25, 2023. Four days after the arrest. Referee Lyonel Norris orders a Rule 20.01 competency evaluation at the earliest procedural opportunity. The case docket at that point contained a complaint, a detention order, and six procedural entries. Nothing else. The question of whether the defendant could understand his own proceedings was raised before any proceedings had occurred.

The order itself is fraudulent at origin. Forensic analysis of the signature on the January 25, 2023 Rule 20.01 order (Referee Norris, signing for Judge Toddrick Barnette) finds it byte-identical to the signatures on six other Rule 20.01 orders against six other defendants — same handwritten approval image, cloned across seven separate cases, issued by more than one officer. Mean cross-correlation 0.5284 against a random-baseline of 0.0104: 50.9× separation. The order that launched three years of incompetency proceedings against Guertin is a reused template with a cloned signature. Full forensic analysis, object-level evidence, and the six matched orders are documented at /1st-rule-20-order/.

The same case would accumulate 246 docket events over the next 33 months. 29 Rule 20 competency events — the 93.7th percentile among the 1,288 criminal cases in the MCRO corpus with any Rule 20 activity. 13 distinct judicial officers. 200 court-filed PDFs totaling 12,077 pages — the most heavily documented case in the entire 4,251-document corpus. Zero dispositions. The competency question would not be answered. It would be re-asked, indefinitely.

Dr. Jill Rogstad

The Authorship Cascade

The Rogstad “evaluation” had three people attached to it. Dr. Casey Boland — a forensic-psychology postdoctoral fellow — conducted the interview. Rogstad’s own report states this on the record:

Rogstad signed the findings. Metadata authorship of the resulting PDF — the Author and Creator fields written automatically by the software that produced the file — was “GuzmanC” (Chela Guzman-Weigart). Three people. Three roles. One evaluation that would anchor three years of incompetency proceedings against a defendant on four firearms charges.

Effective Contact Time

The interview was scheduled for approximately two hours on March 1, 2023, at the Hennepin County Government Center. Rogstad’s own report documents that Guertin arrived thirty minutes late, having called ahead to say he would rather be late and “100% prepared” than arrive on time. Effective clinical contact: approximately 1.5 hours.

The Diagnosis, and What Rogstad Admitted She Could Not Verify

The diagnosis, verbatim:

The expertise limit, verbatim, from the same section as the diagnosis:

The attempt to verify the patent, verbatim:

That date is the cover-page date on the report. The single documented attempt to verify the patent occurred on the day the report classifying the patent as delusional was filed with the court. The patent had been publicly searchable on USPTO.gov for twenty-four days.

The Recommendation

Rogstad did not stop at incompetency. Verbatim, from the final section of her report:

Civil commitment was not downstream. It was recommended in the first report.

The Scorecard

Rogstad classified 11 categories of the defendant’s statements as evidence of delusional thinking. Those categories were independently tested against the MCRO forensic database and the public record:

Four of eleven “delusional” claims (36.4%) have independently verified factual predicates. Four more are partially verified. Nine of eleven have some factual basis. Two are not supported by any forensic record — the assassination-conspiracy claim and the website-symbology claim. Those two are the minority.

The Witness Statement

Three months after Rogstad filed her evaluation, and three weeks before she was scheduled to testify at the evidentiary hearing, she met privately with Assistant Hennepin County Attorney Jacqueline Perez for witness preparation. The memo Perez wrote afterward is in the prosecution’s own first person. It was subsequently filed into the federal court record in Guertin v. Hennepin County, Case 0:24-cv-02646-JRT-DLM, Doc. 43, page 195 of 196.

It opens: “On June 12, 2023, I met with Dr. Jill Rogstad in preparation for upcoming testimony. Below is a non-verbatim summary of our conversation to the best of my recollection.” “I” is the prosecutor. Defense counsel is not named as present. Defense counsel is not referenced as notified.

The memo contains two lines that are load-bearing.

Four months after the patent was granted. On the prosecution’s own letterhead. The evaluator who had classified the patent as delusional told the prosecution, in a private meeting without defense notification, that verified patents would not change her conclusion. The forensic opinion she would deliver under oath at the July 2023 evidentiary hearing had been fixed in advance. The evidence was not the variable. The conclusion was.

The second line, from the same memo:

Place adjacent to the current-date forensic record. The data Guertin “talked about” is this:

Dr. Adam Milz

Milz conducted a two-hour videoconference interview on January 3, 2024. Guertin — by then under a stayed order of civil commitment — recorded the session. The video was later filed publicly in federal court as Exhibit AB, making the interview itself the most independently verifiable forensic evaluation in the MCRO corpus: anyone can compare what was said against what was written.

The sole external collateral contact, verbatim:

Rivers was the same attorney Guertin had spent months alleging was compromised — the same attorney who had told Guertin in May 2023 that he had “very powerful people keeping an eye on” him. That eight-minute call was Milz’s sole contact with any party outside the defendant himself.

On the question of whether the January 2024 evaluation was independent of the March 2023 evaluation, the report is explicit:

The diagnosis was carried forward from Rogstad. Milz added his own characterization of Guertin’s claims — a definition that is useful to note verbatim for the inversion table that follows:

The Substance-Use Fabrication

Milz’s Substance Use History section, verbatim:

Because Guertin recorded the interview, it is possible to compare what Milz attributed to him against what he actually said — line by line, with timestamps. The AI-generated transcript of the Zoom recording was subsequently filed into federal court.

Substance	Transcript of the interview (video filed publicly)	Milz’s report
Methamphetamine	Milz: “Meth, cocaine, pills, anything?” (00:13:34). Guertin: “No.” (00:13:36).	“history of problematic use”
Crack cocaine	Never mentioned by either speaker. Zero occurrences in ~2 hours of transcript.	“history of problematic use”
LSD	Never mentioned by either speaker. Zero occurrences in ~2 hours of transcript.	“history of problematic use”
Marijuana	Milz: “Any weed, anything like that?” (00:13:27). Guertin: “No.” (00:13:30).	“history of problematic use”
DMT	Single transformative 2018 experience, described in detail at 00:11:41. Guertin reported it led him to stop drinking and start exercising.	Listed alongside meth and crack.

Of the five substances Milz attributed to Guertin as “problematic use,” four were either explicitly denied during the recorded interview or never mentioned by either speaker. The fifth was described as a single, transformative experience Guertin credited as the turning point that ended his drinking.

The Self-Contradiction

Two sections of Milz’s January 11, 2024 report on Matthew Guertin. Same document. Same author. One page apart:

The contradiction is Milz’s, not Guertin’s. Both quotes are verbatim from the same report.

The Administrative Order

Five days after Milz filed his report, Referee Danielle Mercurio entered a Finding of Incompetency and Order. Administratively. No evidentiary hearing. No testimony. No opportunity for the defendant to contest the findings. Civil commitment screening triggered on a five-day deadline.

The scorecard on Milz’s 19 classifications:

12 of 19 (63.2%) have independently verified factual predicates. 16 of 19 have some factual basis. Zero are affirmatively contradicted by the evidence. The ten specific entities Milz cited by name as evidence of persecutory paranoia — CIA, Army, Air Force, DIA, DARPA, Indo-Pacific Command, Lockheed Martin, Fox Corporation, Forcepoint, State Department — appear in 1 to 21 distinct court-filed documents each. Ten for ten.

Dr. Katheryn Cranbrook

The Nested Conflict of Interest

Cranbrook’s report documents the origin of her assignment in one paragraph:

The conflict she solved: Milz had been named as a defendant in the federal lawsuit. The conflict she created: Cranbrook is Chief of Psychological Services — Milz’s direct supervisor. The organizational unit she runs produced all three evaluations. She had more institutional exposure to the federal lawsuit than Milz did. The conflict of interest was not resolved by the substitution. It was deepened.

Cranbrook’s report then states, verbatim:

The supervisor maintained her subordinate’s finding — with zero independent clinical data — at the precise moment her subordinate was about to be deposed in federal court.

Zero Hours of Clinical Contact

Cranbrook’s report documents no in-person, video, or phone interview with Guertin. Her entire evaluation rests on record review, email correspondence, and one collateral phone call on July 26, 2024 with a Vail Place case manager. The case manager had no knowledge of his medications.

Guertin did not refuse categorically. He requested his attorney be present — citing, in writing, documented inaccuracies in prior reports. Cranbrook responded that she needed to meet with Guertin, not his attorney. No further communication occurred. The report was filed anyway.

What Cranbrook Documented Against Her Own Diagnosis

From Cranbrook’s own report:

The state’s own civil-commitment system had determined Guertin no longer met the statutory criteria for commitment. Cranbrook’s own report documents this. In the same report, she maintains the diagnosis that had originally triggered the commitment.

The Circular-Reasoning Trio

Three loops. Each is a separate sentence in Cranbrook’s own report. Each classifies a legally protected act as a symptom of the disorder it is used to diagnose.

Loop 1 — Refusing the evaluation is the diagnosis

Guertin’s stated reasons for declining — wanting effective counsel present, citing documented inaccuracies in prior reports, citing fraud in the civil-commitment discovery — are classified as the symptom, not evaluated as legitimate. The act of declining becomes the diagnosis.

Loop 2 — The federal lawsuit is the diagnosis

The allegations in question are filed civil-rights claims in the United States District Court (Guertin v. Hennepin County, 0:24-cv-02646) — a First Amendment Petition Clause act. Filing a lawsuit naming the evaluators as defendants becomes evidence of the illness the evaluators diagnosed.

Loop 3 — Challenging the evaluation is confirmation

Guertin’s written objections — the thing he was supposed to do to contest the prior findings — are themselves the confirmation of the prior findings. The act of challenge becomes evidence that no challenge is possible.

The three loops share one structure: the defendant articulates a factual basis; the evaluator classifies the act of articulating the factual basis as a symptom; the symptom classification reinforces the diagnosis that motivated the evaluation. An unfalsifiable diagnostic system is not a diagnosis. It is a perimeter.

The Treatment Catch-22

The rest of Cranbrook’s Competence-Related Prognosis section — the full mechanism in five verbatim sentences:

Antipsychotics are the prescribed path to competency. The defendant will not take them voluntarily. The statutory threshold for forced medication is not met. Therefore the prognosis is “poor.” The case cannot advance to trial. The defendant cannot be treated against his will. The competency question cannot be resolved.

This is not a clinical dead end. It is an administrative one. Refusal to medicate equals indefinite incompetence. Indefinite incompetence equals indefinite supervision. The statutory frame is designed such that the person declared incompetent holds the only key — and turning the key requires first accepting that the diagnosis is correct. A perpetual-motion machine dressed as a prognosis.

The Single-Source Authorship Collapse

Three “independent” evaluations. Three different signatures. The metadata of the resulting PDFs tells a different story.

Below, the Milz report (filed January 11, 2024) and the Cranbrook report (filed December 20, 2024) compared at the file-metadata level. These fields are written automatically by the software that produced the PDF. They are not set by the person whose name appears on the signature line.

Metadata Field	Milz Report — Jan 11, 2024	Cranbrook Report — Dec 20, 2024	Match
XMP Toolkit	Adobe XMP Core 5.6-c142 87.162696, 2018/09/16-15:11:55	Adobe XMP Core 5.6-c142 87.162696, 2018/09/16-15:11:55	IDENTICAL
Creator	Hines, Anne	Hines, Anne	IDENTICAL
Author	Hines, Anne	Hines, Anne	IDENTICAL
Producer	Microsoft: Print To PDF	Microsoft: Print To PDF	IDENTICAL
PDF Version	1.7	1.7	IDENTICAL

Two reports. Filed 344 days apart. Attributed to two “independent” evaluators. Five metadata fields byte-identical, including an Adobe XMP Core library compiled on September 16, 2018 and carried forward unchanged into both documents. Neither Milz nor Cranbrook appears in any authorship metadata field of their own reports.

The Rogstad report carries a separate authorship anomaly. Its Author and Creator fields contain “GuzmanC” — not Rogstad, not Boland. The Rogstad report is also absent from the MCRO public database (the has_pdf flag on its docket entry is false), meaning it was distributed to the parties but never posted as a publicly downloadable document.

The “fabricated after the fact” defense is foreclosed: the January 2024 Milz metadata was logged into the federal court record on August 7, 2024, four months before the December 2024 Cranbrook report existed. The .eml file transmitting the Cranbrook report passes DKIM, DMARC, SPF, and ARC authentication — genuinely sent from Hennepin County’s Microsoft 365 infrastructure (Azure tenant 8aefdf9f-8780-46bf-8fb7-4c924653a8be).

The Diagnostic Labels Roster

Every diagnostic label asserted across the three reports. Verbatim from the primary sources. In the order they were entered.

What the Reports Called Delusional. What the Record Shows.

Left column: verbatim evaluator language from the three primary-source Rule 20 reports. Right column: what the forensic record independently documents for the same factual predicate. Both sides are on the record. Both sides are verifiable. The juxtaposition is the argument.

The Override: Diagnostic Inversion as Operational Mechanism

On April 3, 2025, Judge William H. Koch found Guertin competent to proceed. It was the first contested competency hearing in the case. Guertin had testified under oath for approximately 40 of 61 transcript pages. Koch’s order was the first MSIP-labeled (Microsoft Information Protection) document in the entire 4,251-document MCRO corpus. Twenty-six days later, it would be effectively set aside.

Donnelly’s Logic Trap

At the March 5, 2025 contested hearing that preceded Koch’s finding, defense counsel Emmett Donnelly instructed the court NOT to enter Guertin’s forensic exhibits (indexes 122–125 — 266 pages of document-manipulation analysis). Per Koch’s order, footnote 1: the defense declined to offer the exhibits. Donnelly also characterized those exhibits as “doctored.”

Name the trap. If the exhibits are doctored, that is a defense accusation — an allegation against whoever doctored them. The natural defense move is to enter them into evidence and make the accusation the centerpiece of the case. Withholding them makes no defensive sense. The only reading under which withholding is rational is the reading where the exhibits are real, and entering them would have damaged the competency framework itself. Donnelly walked the logic trap voluntarily, on the record, at the hearing where his client was found competent.

April 17, 2025 — Signaling to a New Judge

Fourteen days after Koch found Guertin competent, the case was before a different judge. Sarah Hudleston’s first substantive hearing on the case lasted 19 minutes. Donnelly told her, on the record:

Defense counsel, in open court, told the new judge that counsel disagreed with the prior judge’s finding that their client was competent. Donnelly then told Hudleston that defense counsel had not adopted and would not pursue Guertin’s 50-page Motion to Dismiss — a document whose factual claims would later be verified 66.7% true by independent forensic analysis. The prosecution agreed the motion should not be addressed.

April 29, 2025 — The Prosecutor’s Self-Refuting Statement

Twelve days after the April 17 hearing, Hudleston convened a 22-minute proceeding. Under her own “duty of candor” framing, she asked prosecutor Mawerdi Hamid about the discovery-manipulation evidence Guertin had described. Both of the following statements appear on the same page of the April 29 transcript — page 10:

One claim. One retraction. One page. The prosecutor who conducted cross-examination at the March 5 contested hearing — where the discovery evidence was discussed under oath — tells the court on April 29 that the state is not aware of it, then two sentences later confirms the state is aware of it. Both statements on the record. Both from a lawyer under duty of candor. The contradiction is not between two separate moments. It is not a recollection that evolved. It is on one page.

The Motion to Dismiss Trigger

Hudleston ordered a fourth Rule 20.01 competency evaluation. On the record (April 29 transcript, p.17), she identified the basis:

The Motion to Dismiss. The document defense counsel refused to adopt. The document the prosecution agreed should not be addressed. The document whose substance was never argued in open court. Its mere existence triggered a fourth competency evaluation, twenty-six days after Koch had found Guertin competent. The logical limit of the diagnostic inversion: filing substantive legal claims is itself the symptom that justifies not hearing them. The only way to be heard is to stop filing. Stopping filing is no defense. Filing is evidence of incompetence to be heard.

On the Same Record

At the same hearing where she ordered the fourth Rule 20 evaluation over patent-theft claims characterized by three prior evaluators as delusional, Hudleston said this to Guertin (April 29 transcript, p.14):

The judicial record now contains, on the same day, in the same hearing: a competency evaluation order based on patent claims characterized as delusional, and a judicial acknowledgment that the factual predicates of those claims are accurate.

“The Witness” — On the Record, Under Oath

On March 5, 2025, in the contested competency hearing before Judge William H. Koch, Guertin took the stand and was examined. A single paragraph of that testimony is reproduced verbatim below. The density of factual claims, stacked at paragraph speed, is the refutation. You are reading the unedited voice of the defendant the system has classified as unable to rationally participate.

The Witness — Matt Guertin

“Yes, the — What’s brought me here is three Rule 20 exams that have determined I’m incompetent with the initial stayed over civil commitment being completed on my behalf successfully, and that hearing originally took place on August 1st of 2023 and after which point on August 3rd of 2023 I was emailed a PDF of 80 photographs from my court-appointed attorney at the time Michael Biglow. I never asked for these photographs so they didn’t get reviewed until a few months later at which point I noticed discrepancies, and then filed my first ever court motion in my life on January 5th of 2024 which was seeking discovery materials. And that’s kind of been my consistent path and relentless pursuit of the truth ever since. And as a result the — Dr. Michael Robertson who conducted my civil commitment exam hearing on August 1st over a Zoom hearing relied on those 80 photographs that were provided in the manipulated discovery materials that I was given. And then the subsequent two Rule 20 exams not only also used my claim of there being manipulated evidence as evidence of my supposed mental health issues and a reason why I needed to be placed on powerful antipsychotic drugs against my will to make me competent, but also they — there’s false narratives being perpetuated in all subsequent Rule 20 exams after the first one that would include me being very candidly honest in the first one and saying that I’ve used every drug in my life besides heroin, and then in subsequent exams it says basically that it makes me look like I have a continuing and ongoing problem with every drug besides heroin. It says that I have a history of suicide and self-harm in the subsequent reports which is completely not true. I’ve never been hospitalized for suicide. I’ve never attempted suicide. So that’s blatantly untrue. It says that I have a history of psychosis, and that’s untrue, there is no history of psychosis. There’s no documentation that they can provide, that the prosecution can provide that proves that I do have it because that’s not true. There was a letter submitted by my California physician that I’ve been seeing, I’ve been on the same prescription for ten years since 2016 and that my California physician confirmed that there’s no history of psychosis, et cetera. It says that I have a history of mania, that’s untrue.”

On authentication — This transcript was never filed into the state docket and the court-delivered PDF carries an intentionally misdated filename (“March 5, 2024.pdf” for a March 5, 2025 hearing). Authentication is therefore established via the delivery email itself — DKIM-signed, SHA-256-hashed, preserved in the email archive with full headers intact. ► Read the full transcript — opens directly to page 13 ► Delivery email — authentication layer (DKIM + hash)

Three evaluators across twenty-two months. Two metadata pipelines. Approximately three and a half hours of direct clinical contact. Thirty-seven claims classified delusional. Twenty-one independently verified. Zero contradicted by the evidence. A civil-commitment recommendation in the first report. A fabricated substance-use history in the second. A zero-contact “maintained” diagnosis in the third. A circular-reasoning trio that makes the diagnosis unfalsifiable by design. A treatment catch-22 that makes the proceedings indefinite by design. A judge who — on the same record where she orders a fourth evaluation over patent-theft claims — acknowledges the patent was stolen.

This is not a medical finding. It is a containment mechanism. The claims cannot be heard on their merits because the capacity to raise them has been classified as the illness itself.

What remains is to look at what the so-called incompetent did with the three years he was told he could not understand.

What Was Actually Said — On the Record

Eight short clips pulled from the October 7, 2025 Zoom competency-review hearing before Referee Lori Skibbie. The exchanges are between Guertin and his court-appointed public defender, Raissa Carpenter, captured during the hearing and in the hallway conference that followed. They surface two things the on-paper record does not: that the cycle Carpenter describes is structural rather than medical, and that attendance at the evaluation — the single thing every downstream communication now frames as mandatory — is not a condition of release, by the defense attorney’s own on-the-record admission.